Having DNA samples of all or almost all members of a population of, say, a few hundred or thousand individuals, is it possible to draw the entire family tree of those individuals? (Let's not assume monogamy or any other facilitation).

This may be a sum of two different questions:

  • does comparison of two DNA samples give any information other than topographical distance between individuals? In other words, is it possible to tell two sisters from a mother and a daughter by DNA alone?

  • (more a mathematical question than a biological one, maybe) having the relative topographical distance of all members of a group, can their full graph be drawn?

  • 1
    $\begingroup$ Not confident enough to post an answer but I very much doubt it. There is a large element of randomness involved, I am not simply a mixture of my parent's DNA. Processes like chromosomal crossover ensure that a significant part of my genome is original so to speak. Therefore I doubt you can reconstruct a genealogy from this information. $\endgroup$ – terdon Feb 9 '14 at 14:30
  • $\begingroup$ @terdon: recombination is easily modellable. I'm pretty much sure it is possible even with relatively sparse genotypic data, not to talk about whole genomes. $\endgroup$ – alephreish Feb 9 '14 at 21:00
  • $\begingroup$ @har-wradim how can it be modeled? It is a largely stochastic process so I don't see how you could infer my parent's genomes from mine. $\endgroup$ – terdon Feb 9 '14 at 21:02
  • $\begingroup$ @terdon: it's not completely stochastic on the one hand and many stochastic processes are modelled on the other. You are still nearly 50% identical to each one of your parents and this identity covers long stretches of DNA. $\endgroup$ – alephreish Feb 9 '14 at 21:06
  • 1
    $\begingroup$ @terdon: The task is not to infer your parents' genomes - the task is to find them in a set of genomes known beforehand. $\endgroup$ – alephreish Feb 9 '14 at 21:28

No, it is not possible to determine ancestry with just two DNA samples, but it is possible with three: http://www.nature.com/nature/journal/v487/n7407/full/nature11251.html.

Essentially, each sequence region in the progeny can be traced to either the mother or the father, but you really need both sequences to have any confidence. That would get you one node on your tree. You would need additional samples from higher generations to rebuild an entire ancestry.

However, it might be possible with hundreds of samples from laterally related individuals. In fact, I think this is the fundamental hypothesis of the 1000 genome project, http://en.wikipedia.org/wiki/1000_Genomes_Project, and it pretty much worked. Isolated populations can be discriminated by broad DNA sequence analysis.

To reconstruct your tree for a single ancestry, you would just want to use more closely related individuals (many, many of them), and then generate a phylogenetic tree, which should roughly correspond to the family tree you are investigating.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.